platinum films

Katsuya Iwashina , Akihide Iwase , Yun Hau Ng , Rose Amal , and Akihiko Kudo. Journal of the .... Fuel Cell Geared in Reverse: Photocatalytic Hydrogen...
2 downloads 7 Views 2MB Size
J . Am. Chem. SOC.1984, 106, 6537-6542 metabolism of living organisms. It is the prime energy source for most endergonic biological systems and participates in numerous enzymatic reactions. As we presented in our preliminary report,' the HOESY experiment is able to evaluate solvent and intramolecular proton dipolar relaxation for the three phosphates in ATP. We ran the HOESY experiment for 0.5 M ATP in fully deuterated phosphate buffer (100% D 2 0 , repeated lyophilizations). Figure 7 shows individual slices at three phosphorus signals along the F, dimension (proton chemical shift) in the HOESY experiment. Each phosphorus slice has a small peak corresponding to the HOD peak, due to the small number of residual solvent protons in the sample. The y-phosphorus slice gives the highest H O D intensity which may be due to (a) the terminal phosphorus being most accessible to the water and (b) in the buffer (pH 5.6), there still being one proton attached to the y-phosphorus through oxygen (pK,s for y-phosphorus in ATP are 4.1 and 6.525). In the a-phosphorus slice, two intramolecular dipolar interactions are indicated: (a) with H-5' protons, close to the sugar ring, and (b) with the H-8 proton of the base. This is consistent with the anti conformation being populated, which is expected for solutions of A T P . ~ ~ (25) Phillips, R.; Eisenberg, P.; George, P.; Rutman, R. J . Eiol. Chem. 1965, 240, 4393.

(26) Schweizer, M. P.; Beroom, A. D.; Ts'o, P. 0.P.; Hollis, D. P. J . Am. Chem. SOC.1968, 90, 1042.

6537

Conclusions The heteronuclear 2-D N O E experiment should prove to be a powerful albeit sometimes demanding experiment for structural elucidation and conformational analysis, both in 13Cand 31PN M R spectroscopy. It should be noted that this experiment gives information through space (dipolar interaction) rather than through bonds (J coupling). For carbon-13, it provides information about dipolar interactions between quaternary carbons and nearby protons. A disadvantage of the HOESY experiment for relatively weak N M R nuclei (e.g., l5N, 13C)is that in order to detect cross peaks a concentrated sample may be necessary. This is especially true for nonprotonated sp2 or sp- hybridized atoms at high magnetic fields, where NOES may be quite small. The HOESY technique should prove useful for the study of molecular conformation and solventsolute interactions in a variety of chemical and biological systems. Since the observed nucleus is not limited to I3Cor 31P,HOESY experiments are also expected to have significant potential in studies of peptides (15N), organometallics and metalloenzymes8(spin metals), and solvation studies of inorganic spin 'I2ions.

Acknowledgment. We thank Dr. Ad Bax for thoughtful contributions and discussions. This work was supported by N I H Grants RR-01317 and GM-29778 and N S F Grant C H E 8101509. Registry No. ATP, 56-65-5; camphor, 76-22-2; fluoranthene, 206-44-0.

H2 Photoproduction by Nafion/CdS/Pt Films in H20/S2Solutions Albert W.-H. Mau,+ Chorng-Bao Huang, Noriyoshi Kakuta, Allen J. Bard,* Alan Campion, Marye A. Fox, J. Michael White, and Stephen E. Webber Contribution from the Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712. Received March 28, I984

Abstract: We report the construction,characterization, and evaluation of a novel system for photocatalytichydrogen generation that comprises semiconductor (CdS) crystallites embedded in a polymer (Nafion) matrix that also contains a hydrogen-evolution catalyst (Pt). Hydrogen-production efficienciesfrom water containing a sacrificial electron donor (sulfide ion) were greater than those commonly obtained with unsupported colloidal or powdered semiconductors under similar conditions. Several factors dramatically influence performance: the crystalline form and surface composition of the semiconductor and the extent of catalyst dispersion throughout the film.

Photocatalytic and photosynthetic reactions at semiconductor particles provide the possibility of the utilization of solar energy for the promotion of useful chemical reactions.'V2 Photoreactions at these particles are based on the production of electrons and holes (e-h+) upon absorption of light; these can react with species in solution to produce oxidized and reduced forms. These particulate systems are of interest because they are generally easy to fabricate, are potentially inexpensive, and have a large surface area which facilitates efficient light absorption and heterogeneous charge (e- and h+) transfer to solution species. CdS is an interesting semiconductor material that has been studied as an electrode material; many studies of particulate systems involving CdS have also been r e p ~ r t e d . ~ - ' ~ Particle systems have several disadvantages, however. Since they are dispersed in the solvent system, they are inconvenient to use in continuous-flow systems. They may also tend to flocculate and settle out with time. A previous brief comm~nication'~ Permanent address: Division of Applied Organic Chemistry, CSIRO, P.O. Box 4331, Melbourne 3001, Australia.

0002-7863/84/1506-6537$01.50/0

described the production and immobilization of CdS in a polymer (Nafion) matrix and its use for the photocatalyzed reduction of (1) (a) Bard, A. J. J . Photochem. 1979, 10, 59. (b) Bard, A. J. J . Phys. Chem. 1982, 86, 172 and references therein. (2) Gratzel, M. Acc. Chem. Res. 1981, 14, 376. (3) Krasnovsky, A. A.; Brin, G. P.; Luganskaya, A. N.; Nikandrov, V. V. Dokl. Akad. Nauk. SSSR 1979, 249, 896. (4) Frank, S. N.; Bard, A . J. J . Phys. Chem. 1977, 81, 1484. (5) (a) Darwent, J. R.; Porter, G. J. Chem. SOC.,Chem. Commun. 1981, 145. (b) Darwent, J. R. J . Chem. SOC.,Faraday Trans. 2 1981, 77, 1703. (6) (a) Kalyanasundaram, K.; Borgarello, E.; Gratzel, M. Helu. Chim. Acra 1981,64, 362. (b) Borgarello, E.; Kalyanasundaram, K.; Pelizetti, E.; Gratzel, M. Ibid. 1982, 65, 243. (c) Duonghong, D.; Ramsden, J.; Gratzel, M. J . Am. Chem. SOC.1982, 104, 2977. (7) (a) Harbour, J. R.; Hair, M. L. J . Phys. Chem. 1977,81, 1791. (b) Harbour, J. R.; Wolkow, R.; Hair, M. L. Ibid. 1981, 85, 4026. (8) (a) Kuczynski, J.; Thomas, J. K. J. Phys. Chem. 1983.87, 5498. (b) Kuczynski, J.; Thomas, J. K. Chem. Phys. Lett. 1982,88,445. (c) Kuczynski, J. P.; Milosavlejvic, B. H.; Thomas, J. K. J . Phys. Chem. 1984, 88, 980. (9) (a) Matsumura, M.; Hiramoto, M.; Iehara, T.; Tsubomura, H. J. Phys. Chem. 1984, 88, 248. (b) Matsumura, M.; Saho, Y.; Tsubomura, H. Ibid. 1983, 87, 3807.

0 1984 American Chemical Society

6538 J. Am. Chem. Sm.. Vol. 106, No. 22,1984

Mau el al. * ,

28

50

40

30

20

Fi!gure 1. XRD spectra of cubic (curve A) and hexagonal CdS (curve 6). The dashed lines are those of CdS incorporated in the Nafion film and the solid lines are wmmercial CdS powders.

methyl viologen. In this report we discuss the incorporation of platinum as a catalyst in this Nafion/CdS system and the use of this “integrated chemical system- for the photocatalytic production of hydrogen. These polymer-immobilized semiconductor systems have several advantages. The photoactive system is held in place and so is useful in flow systems. The Nafion/CdS system can he removed for more facile analysis of the reaction solution and regeneration of the semiconductor particles. Moreover, the dispersed semiconductor particles do not tend to flocculate or settle. These membrane systems also offer the potential for “vectorial” charge transfer with separation of the oxidized and reduced products. In addition the polymer matrix itself may play a role in the operation of the system, e.g., by its ion-exchange prope.rties, concentrating some solution reactants and rejecting others. Finally, it is a useful substrate for fabrication of a multicomponent system.

Experimental Section 1. Phtiniutiw of the N.fion Membnnc. Nafion (117 DuPont, equivalent weight 1100.180 pm nominal thickness) pfuorinated cation exchange membranes were uscd for all experiments reported h-n. All Nation films were cleaned by boiling in wncentrated HNO, and then washing with boiling distilled water. The homogeneous deposition of Pt into Nafion films was carried out by soaking in a dilute solution of Pt(NHJ1Il (0.1-0.5 mM)“ for scvcral hours. The amount of Pt wmplex incorporated into the Nation t i h was determined by optical absorption spectroswpy and, dcpcnding on the soaking time. was in the range 0.014.6 mg/m’. The Nafion films with incorporated Pt complex were immersed in 0.1 M NaBH, solution overnight. Under these conditions the complex was reduced to produce Pt-metal particles throughout the Nation films. The platinized Nation films were washed two timcs with boiling distilled water or until the water was neutral. The concentration of Pt in the Nation film could be controlled by varying the soaking time in the Ptamplex solution. The inwrporatcd Pt crystallite size was determined by X-ray powder diffraction (XRD) line broadening methods to be ca. 100 A.IJ Note that this homogeneous platinintion always preceded CdS incorporation (see next subsection). The effect of the H S treatment on the Pt will be discussed below. Pt was also deposited onto Nation films offer CdS inmrporation by RF sputtering at 2 X lW2torr under Ar flow with a Materials Research Gnp.Model MRC8620 triple shutter R&D sputtering module. The rate of Pt deposition was calibrated by measuring the optical absorption of Pt sputtered onto glass plates under similar conditions. This calibration provides our estimatcs of Pt film thickncas, since for sputtered Pt the (10) (a) Hcngllein. A. Ber. Bunrmges. Phys. Chcm. 1982, 86, 241. (b) Hcnglcin. A. 3. Phys. Chcm. 1982.86.2291. ( E ) Alfwi, Z.; Bnhramm, D.; Hcnglein, A. Ibid. 1982.86, 4656. ( I 1) Rosactti, R.; Bnu. L. 3. Phys. Chem. 1982,86,4470. (12) Mctcalfc, K.;Heater, R. E. 3. Chem. Soc.. Chent. Comntun. 1983,

.__.

112

(13) Krishnsn. M.;While, J. R.; Fox. M.A.; Bard, A. I. 3. Am. Chrm. Soc. 1983, 105. 7002. Meislacr. D.; M d n g , R.; Kartcning, B. Chem. Phys. Lctf. 1983, %. 34. (14) Dbra, S. C. I d . 3. Chem. 19lO.8, 193. (IS) (a) Adam, C. R.;Bend. H.A,; Curtis. R. M.;Mrhnheimer, R. 0. 3. Cotol. 1962, I, 336. (b) Alae. M.S.; Rouhaai. M.D. 3. Eleefron. M o m . 1979, 8. 289.

Figure 2. SEM photographs of the surface of a-CdS/Nafion and 8CdS/Nafion.

XRD and SEM were of insulfcient intensity to provide an estimate of particle size or thickness. 2 ReP.ntiw of CdSImpregtutfd Naflon Film and Control of the Clyatll Form of CdS. Nation films, with or without inwrporated Pt, were soaked in 0.5 M Cd(NO,), for 1 h and then washed in distilled H,O. The Cd” ion is strongly bound at the anion.sites in the Nafion. The crystal form of CdS (a-hexagonal or @-cubic)in Nafion dcpcnded on the method of preparation. The XRD data for these different CdS forms are presented in Figure 1. As will be described below the crystal structure affected the efficiency of hydrogen production. Metbod I 8-C& (Cubic). The Nafion film was soaked in boiling. dilute HN03 (pH 2.1) and H# gas bubbled for IO min. The films were then washed with boiling H,O. Thesc films are bright orange. and from the XRD widths the average CdS crystal size was estimated as ca. 200

A ( ~ Z OA).15

Method Ik wCdS (tinu@). The Nafion was soaked in distilled HIOand bubbled with H,S gas for IO min, all at mom temperature. The films were then washed with boiling H1O, These films were bright yellow, and the XRD line widths gave an average crystal size of ca. 200 A ( S O A).” We note that in an earlier report with use of CdCI, and Method I P only a-CdS was formed. Scanning electron microscopy (SEM)revealed a striking difference in the surface roughness of a- and BCdS in Nafion (Figure 2). Cross-sectional SEM of CdS/Nafion did not demonstrate any morpho. logical difference in thcsc two crystal forms in the interior of the mcmbrane. X-ray fluorescence detection of Cd demonstrated that Cd was csscntially homogeneously distributed throughout the Nafion, with a slightly higher concentration near the outer surfaces. As will be disolsscd later, the BCdS (cubic) is much more adivc with respect to HI photoproduction than a-CdS (hexagonal). The extreme disparity in the appearance of the surface is the only morphological differencewe have found for thcsc two preparations. One may speculate that the observed difference in activity is a result of surface area at the CdSsolution interface; however, the light that pcnetrates quite deeply into the CdS layer is as photoactive as light absorbed near the surface (see Results and Discussion). The total CdS content of thesc Nafion films was determined by leaching out the CdS with acid (I .O M HCI) followed by polarographic analysis for Cdl+*.The maximum CdS content in the Nation for thesc studies was 1.3 mg/cm’ ( 3 0 . 5 M). Unlike our earlier report.” in the studies reported here the CdS was distributed essentially uniformly throughout the Nation.

J . Am. Chem. SOC.,Vol. 106, No. 22, 1984 6539

H2 Photoproduction by Nafion/CdS/Pt Films 3. Photolysis. For most experiments two different irradiation systems were used: (1) a single lens system with a 1-kW xenon lamp (1.0 W/ cm2) or (2) a multiple station system with a 450-W xenon lamp (0.15 W/cm2). Both used a water filter to remove IR radiation. Light intensities were measured by an EG&G Model 500-1 radiometer. Wavelength studies utilized an argon ion laser (Coherent Radiation Model CR6-UV). This laser has variable output power for the various wavelengths of interest. All the experiments reported herein use 0.1 M Na2S as a sacrificial reagent (pH 13); this prevents or minimizes oxidation of lattice S2- ions in CdS by photogenerated holes6 Most irradiations were carried out at slightly above room temperature, after we determined there to be little temperature effect on the H2 production rates. In most studies the Nafion/CdS membrane (ca. 1 cm X 1 cm) was supported in the bulk 0.1 M Na2S solution. It is also possible to use Nafion/CdS membranes in a “two-cell”configuration, in which the membrane acts as a separator between two different solutions. Results for systems of this type are currently under investigation. For most experiments the irradiated sample was supported in a IO-mL test tube and was immersed in 3 mL of the Na2S solution. All solutions were thoroughly outgassed with N2 before irradiation. Gaseous products were analyzed by gas chromatography (Autoprep A-700) with a 15-ft Porapak Q column at room temperature. To assess the reproducibility of our measurements, simultaneous irradiations of three identically prepared samples were carried out. The relative standard deviation of these measurements was 10%. Quantum yields were measured by using the blue-green lines of the argon ion laser. The power incident on the sample was measured with a laser power meter, and the hydrogen yield was measured as described above. No corrections were made for light absorption or scattering by the film. 4. Reagents. The chemicals used are commercially available: H2S (Linde); nitric acid (MCB); Na2S-9H20(Alfa); CdCI2.2.5H20(Aldrich); Cd(N03)2.4H20(Mallinckrodt);K2PtCI4(Aldrich); H2F’tCI.6H20 (Alfa); a-CdS (Aldrich, 99.999%);@-CdS(Aldrich, 98.88%);and NaBH, (Alfa).

Results and Discussion In our experiments we have identified three primary effects on the rate of photogeneration of H,: (1) CdS crystal structure and the film pretreatment or irradiation history; (2) method of Pt loading and the weight percent of Pt; and (3) excitation wavelength. A subsection is devoted to each of these. In the last subsection we compare the rate of H,production for the Nafion/@-CdS/Pt system with several other CdS/Pt systems. 1. Effect of Crystal Structure and Film Treatment. For most of the results discussed in this subsection the Pt was deposited homogeneously (see Experimental Section) before CdS impregnation. On the basis of smaller set of experiments we believe that the effect of CdS crystal structure is similar with sputtered Pt. One of the most striking findings in our work on Nafion/CdS is the much greater activity of P-CdS (cubic) toward H2 production than a-CdS (hexagonal). The best Nafion/a-CdS system approached two-thirds of the activity of a typical NafionlP-CdS preparation. The NafionlP-CdS preparation also had the advantage of being very reproducible with respect to the rate of H, production, while different Nafion/a-CdS preparations varied by several orders of magnitude. It was verified by XRD that the P-CdS, which is thermodynamically less stable than a-CdS,I6 retained its cubic form after a series of irradiations. N o H2 is produced in the absence of light up to 70 OC. Temperature had little effect on the H2 yield under photolysis over the range 30-50 OC; at 70 O C : the photolysis yield decreased. All the results discussed in detail were obtained within the temperature range 35-45 OC. The rate of H, production always diminished with irradiation time. This loss of activity was not simply the result of consumption of the S2- in solution nor the inner filter effect of the generated S;- ion, since replacement of the irradiated Na2S solution with fresh Na2S did not restore the initial H2 production rate. A grey-blue deposit was observed to build up on the surface of the NafionlP-CdS during irradiation. This deposit could be removed by washing the membrane with boiling H 2 0 , after which the activity was higher than in the initial run. The history of a (16) Ahlburg, H.; Caines, R. J . Phys. Chem. 1962, 66, 185.

Time (hours)

Figure 3. Time dependence of hydrogen production in a 1-cm2Nafion film containing CdS and Pt (0.02 mg/cm2). The integers are successive runs with the same film. Curves 1# and 2# are for a different film

pretreatment. See text for details. particular NafionlP-CdS sample is presented in Figure 3. The percent consumption of NazS in Figure 3 is based on the following assumed stoichiometry: 2hv

+ 2Sz- + 2H,O

CdS

S2,- + H2

+ 20H-

(1)

In a recent publication Aspnes and Heller” demonstrated a strong effect of adsorbed H2 on platinum-group metal-semiconductor contacts. With their results in mind, NafionlP-CdS was pretreated with H2 at room temperature, followed by at least 0.5 h of outgassing with N2, before use. Except for this H, pretreatment this film was identical with the film used to generate curves 1 through 4 in Figure 3. A comparison of a H2-treated (1 #) film to one with no H 2 treatment (1) in Figure 3 illustrates the overall improvement in the Hz production rate for 1 #. In particular the initial rate of H 2 production is improved (there is no hint of an “induction period”), consistent with the Aspnes-Hiller hypothesis that some H2 production is required before contact potentials are favorable for subsequent Hz formation. In Figure 3, 2 # refers to the same film as 1#, after it was washed in boiling H 2 0 and restored to the same NazS solution. The results of H2 pretreatment suggested the following experiment to test the effect of irradiation history on the H, production rate (see Figure 4). A NafionlP-CdS membrane was irradiated for 1 h with the low-power xenon lamp (0.15 W/cm2) yielding an initial H, production rate of 3.9 pmol/(cm2-h), and when the low-power lamp was replaced with the higher-power xenon lamp (1.0 W/cm2) and the membrane irradiated for 6 min and then returned to the low-power system, the H 2 production rate increased by a factor of about 2. Surface analysis by XPS (VG ESCA-5) indicates that the oxidation state of Pt may be altered by this H,treatment. The Pt(4f7,,) binding energy (BE) was 72.1 eV on fresh films. After irradiation in the manner described above, the BE dropped to 71.3 eV, essentially the same as that for a Pt foil (BE = 71.2 eV). We conclude that Pt was partially oxidized in the fresh films and that during the early stages of hydrogen production, it was reduced to Pto. This change in oxidation state is, we believe, related to the change in H 2 production activity. To investigate this effect further, we probed the state of Pt before CdS incorporation into the film. After Pt incorporation and reduction with NaBH4, the Pt(4f7,2) BE indicated that Pt2+ complexes were completely reduced to PtO. Thus the oxidized Pt species is formed at some point during the CdS incorporation process. Thus while not excluding the process proposed by Aspnes and Heller,” we consider the reduction of partially oxidized Pt as responsible, in part, for the observed induction period. (17) Aspnes, D. E.; Heller, A. J . Phys. Chem. 1983,87, 4919.

6540 J . Am. Chem. SOC.,Vol. 106, No. 22, 1984 40

Mau et al.

c i +I

30 c

.-0

I

E"

-

20

c

450

500

550 W a v e l e n g t h (nm)

650

600

Figure 6. Photoaction spectrum of the CdS-Pt (0.04 mg/cm2) film (obtained by laser excitation (50 mw)). The dashed line is the UV 0

2 4 Time (hours)

6

spectrum of the CdS-Pt film.

Figure 4. Hydrogen production of the CdS-Pt (0.04 mg/cm2)-Nafion film by a low-power (0.15 W/cm2) xenon lamp. The dashed line corresponds to a 6-min irradiation by a high-power (1 W/cm2) xenon lamp. Ln

t ll

. N r

450

P t loading ( w t % )

Figure 5. Effect of Pt loading on hydrogen production for different Pt-CdS systems (for 3 h of irradiation). Curve A: chemically reduced Pt:Nafion/P-CdS film. Curve B: photoreduced colloidal Pt-Colloidal

CdS. Curve C: sputtered Pt:Nafion/&CdS film.

2. Effects of Platinum Deposition and Loading. As discussed in the Experimental Section, two modes of platinization were used: (1) chemical reduction, which is presumed to disperse Pt homogeneously in the Nafion preceding CdS impregnation, and (2) Pt sputtering after CdS impregnation. In the second case the H2 production was very inefficient unless the Pt side of the Nafion/CdS membrane was irradiated. Clearly there must exist some inverse dependence of H, yield on the average separation between the site of photon absorption and the Pt catalyst (this point will be discussed further in the next subsection). The H2 yield was greatest for a sputtered Pt coating on the order of 100 A in thickness; at greater thicknesses, the yield declined approximately exponentially with the thickness of the coating. We interpret this as a simple "filter effect" by the Pt metal. For a comparison of different CdS/Pt systems, the H2 yield as a function of Pt loading was measured for NafionlP-CdS with chemically reduced Pt or with sputtered Pt and CdS colloid with photoreduced P t (following the method of Buhler et al.).'* In all cases 1.3 mg of CdS was used with identical irradiation conditions; the results are presented in Figure 5 . For the cases of chemically reduced Pt with Nafionlp-CdS or photoreduced Pt-CdS colloid, the optimum loading was -4 wt %. For sputtered Pt with Nafionlp-CdS, the optimal loading was 1.5 wt %. Note that for Nafionlp-CdS the optimal H, yield for the sputtered Pt is approximately one-third that of chemically deposited Pt. We ascribe this to a simple geometric effect, Le., the average separation between an absorbed photon and the Pt is larger for the "layered" sputtered Pt:P-CdS system. We do not have an explanation for the much stronger decrease in H, yield for Pt > 4 wt % for the CdS colloid system relative to Nafionlp-CdS.

-

500 550 Wavelength (nm)

600

Figure 7. Photoaction spectra of the CdS-Pt (100 8, thick) film (obtained by laser excitation (SO mW)) with irradiation from the Pt side (0) and the bare CdS side (A). The dashed line is the UV absorption of the CdS-Pt film.

3. Effect of Irradiation Wavelength and Intensity. For all the experiments discussed in this subsection an Ar ion laser was used as an excitation source. This excitation source has several emission lines that can be selected in the region of the CdS band gap, and the power of all lines can be adjusted to be the same. In Figure 6 the action spectrum for H 2 production is compared to the NafionlP-CdS (chemically deposited Pt) absorption edge. As expected, the H, yield increases sharply at the band edge (note the absence of H2 production at the 647.1-nm Kr ion laser line). There is a slight decrease in H2 yield at the shorter wavelengths, where the light absorption by the CdS colloids is known to be increasing. This implies that H2 production near the illuminated surface is slightly less efficient than deeper in the Nafionlp-CdS membrane. Analysis of the effect of penetration depth of the radiation is more straightforward for NafionlP-CdS with the sputtered Pt coating (Figure 7). In this case the H2.yield does not rise as rapidly at the band edge as with the chemically deposited Pt and reaches a clear plateau at shorter wavelengths. Also shown in Figure 7 is the wavelength dependence of H2 production with irradiation from the Pt-free side of the NafionlP-CdS. There is no appreciable Hz yield except at the longest wavelength. At this wavelength some light reaches the platinized side of the membrane, although it is much attenuated. These results reemphasize that CdS must be excited in the vicinity of the Pt catalyst. It is interesting to consider the relation between the lightpenetration depth and H, yield for a sputtered Pt coating in more detail. For purposes of discussion we adopt a model in which the rate constant of H, formation from an excitation event a distance x from the CdS-Pt interface, r ( x ) , is a step function, i.e. r ( x ) = ro

r(x) = 0

L x

(2)

J . Am. Chern. SOC.,Vol. 106, No. 22, 1984

H I Photoproduction by NafionlCdSIPt Films

Table I. Comparison of H2Production by Different CdS/Pt Preparations wt O/o H2 rate,' rate with Pt/ system of Pt pmol/(cm*.h) rate Pt free CdS colloid,b photoreduced Pt 2.5 4.0 10 CdS powder,' Pt powder 4.0 1.o 2.5 Nafionlb-CdS, chem-reduced 1.3 4.2 100 Pt Nafionlb-CdS, sputtered Pt 1.5 2.1 33 Nafionla-CdS, chem-reduced 1.3 2.5 100 Pt a Irradiation source: xenon lamp, 0.1 5 W/cm2. *Preparation according to Buhler et CPreparationaccording to Matsumura et

I

40a

f

E

(0

1

0

30-

I

f 2ol 0

~

1 2 3 4 Laser intensity (70-~rnol of photonslhour)

Figure 8. Light-intensity dependence of the rate of hydrogen production in a Nafion film containing CdS and Pt (0.04 mg/cm2) at 514.5 (0)and 457.9 nm ( A ) .

Thus for constant photon flux the rate of H, formation per cm2 of irradiated surface for any wavelength, RA, is related to the fraction of incident light that is absorbed in the region 0 I x 5 L, i.e.

RA = Iexro(l- e-*") = Zexro(l-lO-'kML)

(3)

where cyA is the extinction coefficient in cm-l and I,, is the photon flux per cm2. In the second term M is the molarity and t Athe decadic molar extinction coefficient of the CdS. If eAML >> 1 then R Areaches a constant value, as observed for X < 500 nm. The relative rate of H2 formation for band edge excitation (RAJ to the limiting yield for deep band excitation ( r d e x )yields an equation for L

(4) Taking either our values for e based on the NafionlP-CdS absorption spectrum or published values for CdS colloids,19we estimate L to be in the range 10-30 pm. Thus our results are consistent with extensive charge or energy migration from the point of initial excitation to the CdS-Pt interface. This calculation does assume a homogeneous Pt coating on the outside region of the Nafion/CdS membrane. If there is significant sputtered Pt migration into the Nafion (on the order of 10 pm) then L in the previous equations would correspond to this migration length. However, we have not been able to directly detect Pt in the interior using Ar ion sputtering/XPS techniques. Extensive charge or energy migration could lead to nongeminate electron-hole recombination or excited-state annihilation. This would result in a sublinear dependence of H2 formation rate on light intensity. This does not seem to be the case, since the rate of H, production is linear with laser intensity at either 514.5 or 457.9 nm for NafionlP-CdSlsputtered Pt (see Figure 8).

4. Comparison of NafionlB-CdS with Other Sacrificial CdS:Pt Systems. It is difficult to establish absolute rates of H2 production, because different workers use different irradiation geometries and/or sources (including filtering). We have approached this problem by comparing our Nafionlo-CdS system (with chemically deposited or sputtered Pt) with two other CdS:Pt systems reported in the literature. We have tried to reproduce the method of CdS:Pt preparation based on these published reports and have compared the rate of H, production under identical irradiation conditions. We have also compared the ratio of the H, production rate with and without Pt catalyst, as a kind of internal calibration. These results are given in Table I (also see Figure 4 and the discussion in subsection 2). Our optimum NafionlP-CdS system with chemically deposited Pt is slightly more efficient than the dispersed (18) Buhler, N.; Meier, K.; Reber, J.-F.J . Phys. Chem., in press (19) Ramsden, J. J.; Gratzel, M., preprint.

6541

CdS colloid system of Buhler et al.16 and significantly more efficient than the CdS powder system of Matsumura et Perhaps the most interesting difference between our Nafion/CdS system and colloid or powder systems is the very large enhancement factor induced by the Pt catalyst, Le., the Nafion/CdS system is inefficient for H, production in the absence of Pt catalyst. We note that the number of moles of H2produced with a typical NafionlP-CdS system exceeds the moles of CdS present by a factor greater than 100. Clearly this system does not involve extensive oxidation of the CdS itself. If the sacrificial reagent Na2S is eliminated, only a very small amount of H2 is produced; for the pH range 6-12 under these experimental conditions we estimate the quantum yield of H2 production to be - 5 X lod (Le., 1 / 2 X 5 X 10" mol of H2produced for each mol of photons absorbed).

Conclusions Thus, we have developed a new technique for the support of highly reactive, small dimension semiconductor particles within a polymeric membrane. When the semiconductor particles of this integrated system are metallized to provide a catalytic reduction site, these materials function as effective photoelectrochemical interfaces for photoinduced electron transfer. Upon photolysis of platinized CdS particles supported within Nafion in the presence of a sacrificial electron donor (Na,S), the production of hydrogen gas by water reduction can be observed. The efficiency of hydrogen production by this integrated chemical system is comparable to the best colloidal or powdered CdS preparations that have been reported, under the same experimental conditions. The crystal structure of the photoactive semiconductor particle greatly influences the yield of hydrogen, with cubic &CdS acting as a more active catalyst than hexagonal a-CdS. Also critical for efficient operation of the system was the method of dispersal of platinum on the semiconductor surface. Either homogeneous chemical deposition throughout the membrane or surface coverage by metal vapor deposition could be used to achieve metal-semiconductor contact. A study of the effect of average separation between the photoactive interface and the metallic reduction center implies that either charge or energy migration occurs over an appreciable distance (upwards of 10 km) or that appreciable diffusion of metal into the polymeric membrane during deposition has occurred. The observation of an induction period before evolution of hydrogen begins is consistent with the Aspnes-Helleri' hypothesis requiring surface hydrogenation in order to establish the requisite contact potential for efficient gas production. However, using XPS we have also demonstrated that this induction period may be related to the reduction of Pt to the zerovalent state. These integrated systems seem to be robust, although after prolonged use the rate of hydrogenation decreases. The photoactivity of a used membrane system can be restored by washing with boiling water to remove surface coatings. Stoichiometric consumption of surface-dispersed CdS is clearly not responsible for the decreased hydrogenation rate, for on a molar basis many more equivalents of hydrogen can be evolved from this integrated system than are present as photoactive semiconductor. Efforts are continuing to understand the effect of the semiconductor crystal structure, of the degree of dependence on the

6542

J. Am. Chem. SOC.1984, 106, 6542-6551

mode of dispersal of platinum on the semiconductor particle, and of the optimum geometry in the integrated system on the efficiency of gas evolution.

Acknowledgment. We acknowledge the financial support of this work by the Gas Research Institute (Contract No. 5982260-0756). We thank Dr. M. Krishnan and J. R. White for many

helpful discussions throughout the course of this study and Dr. M. Schmerling for his help with the SEM. S.E.W. thanks J. Ramsden and Professor M. Gratzel for communicating their results prior to publication. Registry No. CdS, 1306-23-6;Pt, 7440-06-4;Na2S, 1313-82-2;H,O, 7732-18-5; H,, 1333-74-0;Nafion, 39464-59-0.

One-Electron Reduction of Nitrobenzenes by a-Hydroxyalkyl Radicals via Addition/Elimination. An Example of an Organic Inner-Sphere Electron-Transfer Reaction V. Jagannadham' and S. Steenken* Contribution from the Max-Planck-Institut fur Strahlenchemie, 0 - 4 3 3 0 Miilheim a.d. Ruhr, West Germany. Received April 17, 1984

Abstract: The reactions in aqueous solution of a-hydroxyalkyl radicals with para-substituted nitrobenzenes were studied by using product analysis, electron spin resonance, and pulse radiolysis techniques. At neutral pH the a-hydroxyalkyl radicals are quantitatively oxidized to yield the corresponding ketones or aldehydes and H', and the nitrobenzenes are reduced to the radical anions. The mechanism of this redox reaction depends strongly on the substituents on the a-hydroxyalkyl radical (the electron donor) and on the nitrobenzene (the electron acceptor). In the case of a-hydroxymethyl radical, the reaction proceeds by addition to the nitro group to produce an alkoxynitroxyl radical which can undergo an OH--catalyzed heterolysis to give formaldehydeand the radical anion of the nitrobenzene. With the a-hydroxyethyl radical, both addition and "electron transfer" take place, the fraction of electron transfer increasing with increasing electron-withdrawing power of the substituent. The nitroxyl-type adducts undergo a spontaneous unimolecular heterolysis to give acetaldehyde, H', and nitrobenzene radical anion. The rate constants k, (from lo6 s-I in water.

The one-electron reduction of nitro compounds has received considerable attention, particularly as a result of the interest in structure-reactivity correlations concerning radicals* and also in view of the application of nitro aromatics as radiosensitizers in r a d i ~ t h e r a p y . ~In order to reduce the nitro compounds, these have often been reacted with a-hydroxyalkyl radicals, which generally behave as potent one-electron donors., However, as shown4 as early as 1968 by ESR measurements, the mechanism of interaction of a-hydroxyalkyl radicals and nitro compounds is more complicated than expected on the basis of simple electron transfer between the reactants. Depending on the nature of the (1) On leave from Department of Chemistry, Sardar Patel College, Secunderabad-500025, India. (2) For reviews see, e.g.: (a) Neta, P. Adu. Phys. Org. Chem. 1976, 12, 2. (b) Henglein, A. Electroamai. Chem. 1976,9, 163. (c) Swallow, A. J. Prog. React. Kinet. 1978, 9, 195. (3) For reviews see, e.g.: (a) Adams, G. E. Adu. Radiat. Chem. 1972, 3, 125. (b) Greenstock, C. L.; Dunlop, I. In "Fast Processes in Radiation Chemistry and Biology", Adams, G. E., Fielden, E. M., Michael, B. D., Eds.; J. Wiley: New York, 1975; p 247. (c) Greenstock, C. L. In 'Chemistry of Amino, Nitroso and Nitro Compounds and their Derivatives", Patai, S., Ed.; J. Wiley: New York, 1982; p 291. (d) Willson, R. L. In 'Free Radicals, Lipid Peroxidation and Cancer", McBrien, D. C. H., Slater, T. F., Eds.; Academic Press: New York, 1982; p 275. (e) Wardman, P. Curr. Top. Radiat. Res. Q 1977, 1 1 , 347. Adams, G. E.; Breccia, A.; Rimondi, C., Eds. 'Mechanism of Action of Radiosensitizers of Hypoxic Cells"; Elsevier: Amsterdam, 1979; Strahlentherapie, Special Vol. 75, p 20. (4) McMillan, M.; Norman, R. 0. C. J. Chem. SOC.B 1968, 590.

0002-7863/84/1506-6542$01.50/0

a-hydroxyalkyl radical and on solvent or pH, nitroxide-type radicals, formed by addition rather than electron transfer, were d e t e ~ t e d . ~These , ~ observations were supported by the results of pulse radiolysis experiments using optical detection,6 and it was concluded4v6that it is the stability of the conjugate carbonium ions' that determines whether the reaction between an a-hydroxyalkyl radical and the nitro compound proceeds by electron transfer or by addition. On the other hand, it has been pointed outs also that the redox potential of the nitro compound may determine the probability of radical addition or electron transfer. However, the dependence of the reaction pattern on the structure of the reactants has not been systematically studied. Furthermore, the potential influence of solvent and of p H has not received the necessary attention, although it was demonstrated4 already in 1968 that, e.g., the adduct of the hydroxymethyl radical to nitrobenzene (5) Nitroxidt-type radicals have also been detected on reaction of nitro

compounds with a-alkoxyalkyl radicals: (a) Janzen, E. G.;Gerlock, J. L. J . Am. Chem. SOC.1969, 91, 3108. (b) Sleight, R. B.; Sutcliffe, L. H. Trans.

Faraday SOC.1971,67,2195. (c) Wong, S. K.; Wan, J. K. S. Can. J . Chem. 1973, 51, 753. (d) Sancier, K. M. Radiat. Res. 1980,81, 487. ( e ) Reference 4. Studies on a-hydroxyalkyl radicals in nonaqueous solutions are the following: ( f ) Johnson, D. W.; Salmon, G. A. J . Chem. SOC.,Faraday Trans. 1 1977, 73, 256; 1979, 75, 446. (6) Adams, G.E.; Willson, R. L. J . Chem. SOC.,Faraday Trans. 2 1973, 69, 719. (7) Edge, D. J.; Norman, R. 0. C. J . Chem. SOC.B 1970, 1083. (8) Wardman, P. Internal. J . Radiat. Bioi. 1975, 28, 585.

0 1984 American Chemical Society